On Tue, 4 Jul 2000, Alan Cox wrote:
> > I run SUSE 6.4 (kernel 2.2.14 with SMP). My two processors are listed, and
> > I can make SMP on them. But...
> >
> > The problem is that if I set the dma on the ide hard disks using
> > hdparm -d1 /dev/hda
> >
> > and then copy a large size pack of files (say 700Mb) between the sda and
> > the hda, then I have "Segmentation faults" during the copy, or even
> > worse the "cp" command is listed in the processes but nothing works (no
>
> If you instead use a standard 2.2.16 kernel does the thing work (I want to
> eliminate reiserfs and the ide changes from the question first). Do you ever
> see problems under load (eg kernel building) as well with DMA off ?
>
>
I haven't tried the 2.2.16 yet, because I want one with reiserfs support,
so I have to patch it, and I haven't done that yet. Or is there an easy
way with SUSE 6.4 ?
As for the second question, yes, I have discovered that my HD had some
errors when working alone (the ide are connected, but they don't do
anything because the action takes place on the SCSI) :
* scsi : aborting command due to timeout
then
* scsi bus is being reset
and that happened under heavy memory/disk load from a Fourier Transform.
It didn't crash the system, but produced considerable idle periods where
the mouse and keyboard did not answer. Then big noisies from the SCSI disk
where everything takes place, then idle, then...
I tried to set the bios of the adaptec 29160 so that the drive was set at
80Mb/s, but that is worse ; when I make the same test, the system is
totally blocked.
So I guess it's not the dma only... it's dma mixed with a problem with my
adapter/hard-disk.
NOW A BIG QUESTION/PROBLEM of losing data (reiserfs ?) :
Let us look at the exchange of data between SCSI and IDE set to
UDMA :
If I set the dma, and if I make copies from ide to scsi (works on the
other way too) and vice versa, and say I copy 3 files, each time checking
that the file, after having been copied, exists in its destination, and
say that the third copy produces another error I have which is the :
* unexpected IRQ vector 217 on CPU#0 (this error doesn't happen when
working only on the SCSI, as, for example, with the Fourier transform)
followed by the
* SCSI : aborting command due to timeout
Then I have to reboot by pressing reset, and then the two files I had
copied and checked do not exist any more !!!!!!!!!!!! The drive has no
trace of their existence !
I'm lost.
Maybe if I compile the 2.2.16 and use the reiserfs patch... But I'm quite
depressed now. No sleep, no good.
Yann
------------------------------------------------------------------------
Yann Le Du E-mail: [EMAIL PROTECTED]
Theoretical Physics Web : http://cdfinfo.in2p3.fr/~ledu/
1, Keble Road
University of Oxford
Oxford, OX1 3NP Phone : (44) (0)1865 273 989
United Kingdom Fax : (44) (0)1865 273 947
------------------------------------------------------------------------
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/dmentre/smp-howto/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
